Published in:

Advances in Cryogenic Engineering

51 (2006) pp. 59-67.

INERTANCE TUBE OPTIMIZATION FOR

PULSE TUBE REFRIGERATORS∗

Ray Radebaugh1, M. Lewis1, E. Luo2,

J. M. Pfotenhauer3, G. F. Nellis3, and L. A. Schunk3
1
National Institute of Standards and Technology
Boulder, Colorado, 80305, USA
2
Technical Institute of Physics and Chemistry, CAS
Beijing, 100080, China
3
University of Wisconsin-Madison
Madison, Wisconsin, 53706, USA

ABSTRACT

The efficiency of regenerative refrigerators is generally maximized when the pressure

and flow are in phase near the midpoint of the regenerator. Such a phase relationship
minimizes the amplitude of the mass flow for a given acoustic power flow through the
regenerator. To achieve this phase relationship in a pulse tube refrigerator requires that the
flow at the warm end of the pulse tube lag the pressure by about 60 degrees. The inertance
tube allows for the flow to lag the pressure, but such a large phase shift is only possible
with relatively large acoustic power flows. In small pulse tube cryocoolers the efficiency is
improved by maximizing the phase shift in the inertance tube. This paper describes a
simple transmission line model of the inertance tube, which is used to find the maximum
phase shift and the corresponding diameter and length of the optimized inertance tube.
Acoustic power flows between 1 and 100 W are considered in this study, though the model
may be valid for larger systems as well. For large systems the model can be used to find
the minimum reservoir volume that in combination with the inertance tube provides a
phase shift of 60 degrees. This transmission line model is compared with some
experimental results on a small-diameter inertance tube and found to agree quite well
provided some heat transfer is taken into account. Design graphs for a frequency of 60 Hz
and an average pressure of 2.5 MPa are presented for different pressure ratios and for both
adiabatic and isothermal conditions.
∗
Contribution of NIST, not subject to copyright in the US.
INTRODUCTION

Flow in Regenerators

Most losses in pulse tube refrigerators occur in the regenerator. Significant losses also
occur inside the pulse tube component if secondary flows, such as turbulence, acoustic
streaming, and DC flows are not eliminated or minimized. The function of the regenerator
is to transmit a given acoustic or PV power from the compressor to the cold end with a
minimum of losses. These losses include thermal ineffectiveness, lost power associated
with pressure drop, and axial thermal conduction. The first two scale directly with the
amplitude of mass or volume flow through the regenerator. The time-averaged acoustic
power transmitted through the regenerator is given by [1]
m& P
W& = PV& = (1 / 2) P V& cos θ = (1 / 2) P cos θ = (1 / 2) RT0 m& cos θ , (1)
t t ρ0 P0
where the bold symbols represent time-varying or phasor variables, such as pressure,
volume flow rate, and mass flow rate. The phase angle θ is the phase between pressure and
flow. The other symbols are: ρ0 is the average density, R is the gas constant per unit mass,
T0 is the average temperature, and P0 is the average pressure. The expression to the right of
the first equal sign is the general expression, the expression to the right of the second equal
sign applies to a harmonic approximation, the expression to the right of the third equal sign
substitutes the mass flow for the volume flow, and the expression to the right of the fourth
equal sign applies to an ideal gas. From equation (1) we see that the acoustic power flow is
proportional to the component of the mass flow in phase with the pressure. Thus,
minimizing the ratio of the regenerator losses to the acoustic power flow entails
minimizing θ, the phase between flow and pressure. Because of the gas volume in the
regenerator the conservation of mass requires that the flow at the warm end will necessarily
lead the flow at the cold end. In that case the optimum phase relationship is that in which θ
= 0 at approximately the midpoint of the regenerator. Typically the flow at the cold end
will lag the pressure by about 30°, whereas at the warm end the flow will lead the pressure
by about 30°. Such a phase relationship is easily achieved in a Stirling refrigerator by
selecting the appropriate displacer swept volume and phase. Figure 1 shows a phasor
representation of mass conservation in a Stirling refrigerator that shows the optimum phase
relationship of θ = 0 at approximately the regenerator midpoint. In a pulse tube refrigerator
with only an orifice or other resistive flow element for setting the phase, the flow will be in
phase (θ = 0) at the orifice, but not at the regenerator midpoint. Figure 2 shows the phasor
representation of the orifice pulse tube refrigerator (OPTR). In that case θ may be
approximately 30°at the cold end, 50° or more at the midpoint, and as much as 70° at the
warm end. With such a large phase angle the amplitude of mass flow required at the warm
end of the regenerator for a given acoustic power flow will be more than twice that in
which θ = 0 at the midpoint of the regenerator. Such a high mass flow leads to a large
regenerator ineffectiveness and pressure drop.

Regenerator and Pulse Tube Losses

Figure 3 shows how regenerator losses vary with the phase angle between the flow
and the pressure for a given entropy flow at the warm end of the regenerator. Fixing the
entropy flow there is similar to fixing the acoustic power flow at the warm end. The results
 FIGURE 1. Phasor diagram of mass
conservation in a Stirling refrigerator. FIGURE 2. Phasor diagram for an
orifice pulse tube refrigerator.

in this figure were calculated using the NIST numerical regenerator model known as
REGEN3.2 [1]. This figure shows that the minimum regenerator loss occurs with the flow
lagging the pressure at the cold end by about 30°, which typically is achieved with a
Stirling refrigerator. With an OPTR the phase angle is about +30° (flow leading pressure)
at the cold end. As this figure shows, the sum of the ineffectiveness loss S& reg and the
pressure drop loss S& for the OPTR is about twice that for the optimum phase of about -
ΛP
30° at the cold end. The addition of a secondary orifice in the double inlet pulse tube
refrigerator (DIPTR) [2] permits shifting the phase at the cold end to about 0° without a
large additional loss in the secondary orifice. As Fig. 3 shows, the DIPTR then has a
smaller total regenerator loss than the OPTR. However the DIPTR has the disadvantage

FIGURE 3. Regenerator losses associated with ineffectiveness and pressure drop for various phase
angles by which flow leads pressure. Results from calculations with REGEN3.2 [1].
that it opens up a continuous flow loop that allows a DC flow to occur with an
accompanying loss and instability that are not considered in the results shown in Fig. 3.
A relatively new passive flow element, known as the inertance tube, utilizes the inertia
of the oscillating flow to bring about a shift in the phase between the flow and the pressure
[3]. The location of this inertance tube is shown in Fig. 4. For small pulse tube
refrigerators where the mass flow in the inertance tube is small the phase shift may be only
a few degrees. In order to achieve the optimum phase angle of flow lagging pressure by
about 30° at the cold end means that the inertance tube will need to lag the flow at its inlet
(junction with the pulse tube warm end) by about 60° to accommodate the change in phase
through the pulse tube. Such a large negative phase for flow in the pulse tube component
produces a large magnitude of flow for a given acoustic power flow. However, because the
pulse tube is an open tube, large flows there do not contribute to large losses. However, a
secondary flow known as acoustic streaming can occur inside the pulse tube that causes a
loss. Tapering the pulse tube has been shown to reduce this loss [4]. The appropriate taper
angle depends on the phase between the pressure and the flow. For the case of flow
lagging pressure by about 50° at the pulse tube midpoint, the calculated taper becomes zero
[4]. Coincidently, such a phase angle is consistent with having the flow at the inertance
tube inlet lag the pressure by about 60°, which we have already discussed is about the
optimum phase to minimize the loss within the regenerator. Because a 60° phase shift is
difficult to obtain in the inertance tube of small pulse tube refrigerators, we discuss in the
rest of the paper how to use a simple, but relatively accurate, model of inertance tubes to
find the unique diameter and length that gives the maximum phase shift.

INERTANCE TUBE MODEL

Transmission Line Analogy

The equations governing acoustic systems are the same form as those for AC electrical
systems whenever pressure is substituted for voltage and volume flow substituted for
current. The continuous nature of the inertance tube makes it analogous to an electrical
transmission line. The electrical elements of resistance, inductance, and capacitance are
substituted with the fluid elements of resistance, inertance, and compliance. Figure 5
represents a circuit diagram of a small differential element of the transmission line. The
complex impedance of an electrical element is defined as
(Electrical) Z = E / I , (2)
where E is the voltage and I is the current. For fluid systems the impedance is
(Fluid) Z = P / V& = ρ 0 P / m& = ρ 0 Z m , (3)
where P is the pressure, V& is the volume flow rate, and Zm is the impedance to mass flow.
We like the use of Zm because mass flow is conserved, whereas volume flow is not. The
resulting compliance of volume elements are like those used in the phasor diagrams of
Figs. 1 and 2. For the fluid transmission line the resistance, inertance, and compliance per

FIGURE 5. Equivalent circuit of a

FIGURE 4. Schematic of a pulse tube refrigerator with an differential element of a transmission line.
inertance tube and secondary orifice.
unit length appropriate to Zm are:
 32 f r m& 
(resistance/length) r ( D) = (2 / π )  2 5
, (4)
 π ρ 0 D 
(inertance/length) l( D) = 4 /(πD 2 ), (5)
(compliance/length) c( D ) = (πD 2 ) /(4γRT0 ), (6)
where D is the tube inner diameter, fr is the Fanning friction factor, and γ is the ratio of
specific heats. The inclusion of γ in equation (6) implies an adiabatic process. For an
isothermal process we make γ = 1.0 instead of 1.667. For some intermediate heat transfer
process we can take γ to have any value between 1.0 and 1.667. The incorporation of the
friction factor in equation (4) allows the use of this model in both the laminar and turbulent
flow regions. We also take into account surface roughness in the calculation of the friction
factor, but steady-state correlations for tubes are used rather than any complex oscillating
flow correlation. There certainly is some question whether such a steady-state correlation is
valid for oscillating flow, although our comparisons of the model with experiments in a
later section show good agreement.
The reservoir of volume Vr at the end of the inertance tube introduces a compliant load
on the end of the fluid transmission line of
C r = Vr /(γRTr ). (7)
The impedance of this compliant load is
Z r = 1 /(iωC r ). (8)
The complex impedance of a terminated transmission line of length L is given by [5]
 Z + Z 0 ( D) tanh[k ( D)( L − x)]
Z m ( D, x ) = Z 0 ( D )  r , (9)
 Z 0 ( D) + Z r tanh[k ( D)( L − x)]
where the complex characteristic impedance is
r ( D) + iωl( D)
Z 0 ( D) = , (10)
iω c ( D )
and the complex propagation function or wave number is
k ( D) = [r ( D) + iωl( D)]iωc( D). (11)
The convention used here for the position x is that it is zero at the inlet to the inertance tube
(pulse tube end) and increases to L at the reservoir end. The solution to equation (9) for x =
0 gives the amplitude Z m and phase θZ of the impedance at the inlet to the inertance tube.
The sign of θZ is opposite of that used earlier in this paper for θ since θZ indicates the phase
by which P leads m& .

Comparison of Model with Experiment

Measurements of inertance tube impedance were carried out using a commercial linear
compressor to provide the oscillating flow at the inlet to the inertance tube. Flow at the
inlet was measured by a hot-wire anemometer and the flow at the reservoir end was
determined by measuring the pressure in the reservoir and using the ideal gas equation for
an adiabatic process
m& r = P& Vr / γRTr . (12)
The hot-wire anemometer was calibrated against m& r when the inertance tube was replaced
by a needle valve with small volume. The calibration did not appear to be affected by the
 pressure amplitude, which could be varied by adjusting the valve setting. The experimental
inertance tube was made rather long to provide a large compliance component that could be
compared easily with the model prediction. The tube had an inner diameter of 1.58 mm
and a length of 3.60 m. The reservoir volume was 77 cm3. For the measurements reported
here the average pressure was 2.5 MPa, the frequency was 50 Hz, and the average
temperature was about 300 K. With those conditions the thermal penetration depth δt in
the helium gas was 0.219 mm, which yields a ratio R/δt = 3.60, where R is the tube radius.
With such a ratio we expect processes within the inertance tube to be somewhere between
isothermal and adiabatic. Power into the compressor was varied in such a manner to
provide pressure ratios at the inlet to the inertance tube in the range of 1.05 to 1.40. A
comparison between the measured and calculated phase θZ of the impedance is shown in
Fig. 6. A comparison of the measured and calculated flow at the inlet is given in Fig. 7.
Figure 8 compares the flow at the reservoir and Fig 9 compares the phase difference
between the flows at each end. These figures show that there is good agreement between
the model and experimental results. Most of the experimental results lie between the
adiabatic and isothermal curves from the model, which would be expected for the tube
diameter used in these experiments. Because of this good agreement between the model
and experiments, we feel that design curves derived from the model would be useful.

OPTIMIZATION PROCEDURE

Selection of Input Parameters

In the design of pulse tube refrigerators the net refrigeration power is almost always a
fixed input parameter. That refrigeration power is equal to the acoustic or PV power flow
through the pulse tube minus the losses. These losses would normally be calculated in the
design of the refrigerator. Thus, the acoustic power flow into the inertance tube is usually a
fixed and known quantity. For a well-designed pulse tube refrigerator the net refrigeration
power will be about 30% of the acoustic power in the pulse tube for a cold temperature of
80 K. The average pressure and pressure ratio are often fixed for the design, or they are
optimized in some manner. The pressure amplitude then is usually considered a known
input parameter. The design of the inertance tube must be such that for the given pressure
amplitude at the inlet the desired acoustic power flow into the inertance tube occurs.
Equation (1) gives the expression for the acoustic power flow, but includes the mass flow.
From equation (3) we use the expression m& = P / Z m to substitute for m& in equation (1).

FIGURE 6. Measured and calculated phase of FIGURE 7. Measured and calculated mass flow rate
impedance of the inertance tube. at inlet to inertance tube.
FIGURE 8. Measured and calculated mass flow FIGURE 9. Measured and calculated phase
rate at reservoir end of the inertance tube. between flows at each end of the inertance tube.

We then arrive at the equation

2
Zm (1 / 2) RT0 P
= , (13)
cosθ Z P W&
0 t
which defines the locus of points where the complex Zm allows for the specified acoustic
power flow with the given pressure amplitude at the inlet.

Optimum geometry

The locus of points in the complex Zm plane that satisfies equation (13) defines a locus
of points in the D, L plane that give the desired Zm. In general there are several curves in
the D, L plane that represent solutions. Some have negative values of θZ , which we rule
out, and others occur at longer lengths that represent higher multiples of λ/2, where λ is the
acoustic wavelength. In the case of small acoustic power flow we search the locus of
points in the D, L plane that provides the maximum positive value of θZ (negative θ). If
this phase reaches about 60°, we then change our optimization procedure and now reduce
the size of the reservoir volume to keep the phase shift at about 60°. However, here we
only consider small acoustic power flows where we always maximize the phase shift.

OPTIMIZED DESIGN CURVES

Figure 10 shows the maximum phase shift θZ that any single inertance tube can provide
according to the transmission line model discussed here. These results are for a frequency
of 60 Hz, and average pressure of 2.5 MPa, an average temperature of 300 K, a surface
roughness of 1 µm, and an infinite reservoir volume. The effect of surface roughness is
quite small. Curves are given for two pressure ratios and for both adiabatic and isothermal
behavior in the inertance tube. Figure 11 gives the inside diameter of the optimized
inertance tube, and Fig. 12 gives the corresponding length. We note that for small acoustic
powers (less than about 5 W) the radius of the tube approaches the thermal penetration
depth. In such a case the isothermal approximation may be more appropriate. Figure 13
shows the phase difference between flows at the two ends of the optimized inertance tube.
This phase difference is caused by the compliance component in equation (6) of the
inertance tube impedance. This large phase difference shows that the compliance
component cannot be neglected.
 FIGURE 10. Maximum phase shift in inertance FIGURE 11. Inside diameter of optimized inertance
tubes optimized for a given acoustic power. tubes.

CONCLUSIONS

We have shown that the optimum phase between flow and pressure in any regenerative
cycle is that with the flow and pressure in phase at about the midpoint of the regenerator.
Such a phase minimizes regenerator losses. For a pulse tube refrigerator the gas volumes
in the regenerator and in the pulse tube then require that the flow lag the pressure at the
warm end of the pulse tube by about 60° in order to achieve the desired phase within the
regenerator. An inertance tube will provide phase shifts in the right direction, but for low-
power systems the magnitude of the phase shift is less than the optimum. As an aid to the
designer we have shown how a simple transmission-line model of the inertance tube, which
agrees well with experiments, can be used to find the optimum diameter and length of the
inertance tube that gives the maximum phase shift for a given acoustic power.

REFERENCES
1. Radebaugh, R., “Pulse Tube Cryocoolers,” in Low Temperature and Cryogenic Refrigeration, edited by
S. Kakac, H.F. Smirnov, and M.R. Avelino, Kluwer Academic Publishers, Dordrecht, 2003, pp. 415-434.
2. Zhu, S., Wu, P., and Chen, Z., “Double Inlet Pulse Tube Refrigerators: An Important Improvement,”
Cryogenics 30, 514 (1990).
3. Radebaugh, R., “Advances in Cryocoolers,” in Proc. 16th Int’l Cryogenic Eng. Conf./Int’l Cryogenic
Materials Conf., Elsevier Science, Oxford (1997) pp. 33-44.
4. Olson, J.R., and Swift, G.W., “Acoustic Streaming in Pulse Tube Refrigerators: Tapered Pulse Tubes,”
Cryogenics 37, 769-776 (1997).
5. Skilling, H.H., Electric Transmission Lines, McGraw-Hill, New York (1951).

FIGURE 12. Length of optimized inertance tubes. FIGURE 13. Phase shift between flows at each end
of optimized inertance tubes.